I just got the GigaByte X570 AURUS XTREME board a couple of days ago.

Before this I had (and have) the GigaByte X470 AURUS Gaming 7 WiFi and I have discussed my experience with this board and the 3600X I have extensively in the thread https://forums.evga.com/D...-m3001694.aspx#3003763

Running my 3600X (which I got to gain experience with the Ryzen 3000 series CPUs) I had no problems to speak of on the X470 Board. It was running 1.0.0.3 ABBA BIOS.

When I got the X570 board I flashed the backup BIOS to 1.0.0.3 ABBA (F5b) and the main BIOS to 1.0.0.4 B (F10A)

On the newest BIOS I have had nothing but grief trying to configure the BIOS, even something as innocuous as setting the XMP profile causes the board to go nuts.

I switched over to the secondary BIOS with 1.0.0.3 ABBA BIOS and have no problems with the RAM whatsoever.

The RAM I have is four DIMMS of Team Group Edition TDPGD416G3600HC16ADC01 3600 16-16-16-36 and had absolutely no issues with it - even overclocking it to 3733 16-16-16-32 - on the X470 board.

I looked at the GigaByte site and found others who seem to have problems since "upgrading" their BIOS to the new 1.0.0.4 B version and I have to agree with those who say it is borked.

Whenever I set the XMP profile of the RAM on 1.0.0.4 B, the board goes into a round of what I assume is RAM training when it reboots and at some point just hangs, BIOS error 22 is displayed quite often, but not always.

The problem is that I will be getting the 3950X CPU when it comes out at the end of this month and a pre-requisite for running this CPU according to AMD is the BIOS version 1.0.0.4 B.

You can set whatever you want, and it may appear to work, until you try using the system.

Then all kinds of weird and wonderful things happen - like the M.2 boot drive no longer being recognised, to the Windows File Explorer crashing - after which a reboot will reset the BIOS, but of course not the changes made in the AMD portions of the BIOS which, even when you set them back to default will still cause three beeps on reboot (RAM error) unless you run your RAM at the default 2400.

Setting anything in Ryzen Master will also cause the board to crap itself in 1.0.0.4 B
 
I have been experimenting with it for two days and considering I have no problems whatsoever with 1.0.0.3 ABBA and also the fact that I have a clean installation of Windows 10 Enterprise with no additional stuff installed, I can most certainly say that 1.0.0.4 B is FUBAR.
 
The thing is that AMD supplies the AGESA as a "black box" binary and not as a source-code to the motherboard manufacturers so there is very little they can do to correct errors coming from AMD.

I would also reset BIOS to default  ,then install these chipset drivers ,restart &  then re-install the F10A Bios upgrade using the Qflash
 
https://www.amd.com/en/support/chipsets/amd-socket-am4/x570
 
 
Just as a preemptive step anyways ...
 

aka_STEVE_b
I would also reset BIOS to default  ,then install these chipset drivers ,restart &  then re-install the F10A Bios upgrade using the Qflash
 
https://www.amd.com/en/support/chipsets/amd-socket-am4/x570
 
 
Just as a preemptive step anyways ...
 

I've only been a computer techie for 37 years, that just never occurred to me, how could I be such a n00b.

Nec_V20
The 1.0.0.4B is so bad that there are rumours that AMD collaborated with Rockstar Games in the creation of their latest AGESA version

Lol that's funny

Further investigation (I don't get free stuff from AMD) has led me to surmise, that the problem I have experienced with my one chiplet 3600X (and my friend letting me use TeamViewer to manipulate his 3900X) is that the problem only affects one chiplet CPUs.

Further it appears that the problems are RAM related. Namely setting the RAM speed to 3200 (or in Ryzen Master MCLK and FCLK to 1600) results in the system booting normally, but the totally clean installation of Windows10 Enterprise (1903) I have still reacts incredibly sluggishly and File Explorer has great difficulty accessing my M.2 Storage but at least doesn't crash it as it did with the RAM set to the XMP profile of 3600.
 
It takes around three minutes for CineBench R20 to load although the results of running the benchmark are in accordance with the RAM speed I have set.

The one and only time my 3600X system runs as it should with AGESA 1.0.0.4 B is at the stock RAM speed of 2400 MHz.

Working via TeamViewer with my friend's computer shows no problems with the RAM set to 3600 and the system behaves as it should and I cannot find any obvious problems and he has not reported any to me since applying AGESA 1.0.0.4 B

Addendum:

I spoke too soon, even at everything set to Default in the BIOS (RAM 2400), the File Explorer still reacts sluggishly, CineBench R20 will load, but crashes when trying to run a benchmark although CineBench R15 when it loads will run the benchmarks, although the results are totally rubbish compared to what I get when switching back to AGESA 1.0.0.3 ABBA and optimising my system.